An efficient numerical approach to modeling the effects of particle shape on rubble-pile dynamics

Abstract

We present an approach for the inclusion of non-spherical constituents in high-resolution N-body discrete element method (DEM) simulations. We use aggregates composed of bonded spheres to model non-spherical components. Though the method may be applied more generally, we detail our implementation in the existing N-body code pkdgrav. It has long been acknowledged that non-spherical grains confer additional shear strength and resistance to flow when compared with spheres. As a result, we expect that rubble-pile asteroids will also exhibit these properties and may behave differently than comparable rubble piles composed of idealized spheres. Since spherical particles avoid some significant technical challenges, most DEM gravity codes have used only spherical particles, or have been confined to relatively low resolutions. We also discuss the work that has gone into improving performance with non-spherical grains, building on pkdgrav's existing leading-edge computational efficiency among DEM gravity codes. This allows for the addition of non-spherical shapes while maintaining the efficiencies afforded by pkdgrav's tree implementation and parallelization. As a test, we simulated the gravitational collapse of 25,000 non-spherical bodies in parallel. In this case, the efficiency improvements allowed for an increase in speed by nearly a factor of three when compared with the naive implementation. Without these enhancements, large runs with non-spherical components would remain prohibitively expensive. Finally, we present the results of several small-scale tests: spinup due to the YORP effect, tidal encounters, and the Brazil-nut Effect. In all cases, we find that the inclusion of non-spherical constituents has a measurable impact on simulation outcomes.